Lesion detectability and quantification in PET/CT oncological studies by Monte Carlo simulations

In this work, the limits of lesion detectability and quantification accuracy for a state-of-the-art integrated PET/CT system were assessed by Monte Carlo (MC) simulations of realistic anthropomorphic phantoms and patient-derived phantoms, representative of oncological PET /sup 18/F-FDG studies. The dependence of lesion detectability on lesion size and system spatial resolution, lesion-to-background radioactivity concentration ratio (L/B), counts statistics, attenuation and scatter noise, radiation attenuation and scatter correction, data processing, reconstruction algorithms and non-homogeneity of background tissues was investigated. Recovery coefficients were extracted from simulated data and can be used for a MC-based partial volume effect correction method.